Hi There - I’m running a prisma server on Heroku and am getting a ton of timeout errors. I asked Heroku support, and included their response below - I don’t really know where to look for this. I actually took all ws support out of my graphql-yoga server so shouldn’t be asking for subscriptions at all.
This error is causing my app to crash by running the server out of memory on a regular basis. How do I troubleshoot this?
Thanks for reaching out with your issue here.
I’m seeing that Heroku’s router is properly routing requests to your app’s websocket endpoint. However, I’m seeing the following log show up from your web process:
2019-06-18T11:32:50.101858+00:00 app[web.1]: Push to Prometheus Gateway failed with: 2019-06-18T11:32:50.102242+00:00 app[web.1]: akka.stream.scaladsl.TcpIdleTimeoutException: TCP idle-timeout encountered on connection to [metrics-eu1.prisma.io:443], no bytes passed in the last 30 seconds
It would seem like you’re not passing consistent data to Prisma or not properly closing a connection to the service once your operation is complete. Do you mind looking at that endpoint and checking that either of these things are happening?
Friday 14th June 2019
I’m using Prisma which has an automatic deploy to heroku. It should generate a web socket at this url
But I’m just getting an error when I test it, and I’m getting a ton of timeouts / 503 errors I believe because web socket requests are just being ignored or … not sure.